Merge pull request #693 from wren-lang/faster-bit-hash

Fix horrendously bad bit hashing function.

src/vm/wren_value.c

@@ -356,24 +356,18 @@ ObjMap* wrenNewMap(WrenVM* vm)
   return map;
 }
 
-static inline uint32_t hashBits(DoubleBits bits)
+static inline uint32_t hashBits(uint64_t hash)
 {
-  uint32_t result = bits.bits32[0] ^ bits.bits32[1];
+  // From v8's ComputeLongHash() which in turn cites:
-  
+  // Thomas Wang, Integer Hash Functions.
-  // Slosh the bits around some. Due to the way doubles are represented, small
+  // http://www.concentric.net/~Ttwang/tech/inthash.htm
-  // integers will have most of low bits of the double respresentation set to
+  hash = ~hash + (hash << 18);  // hash = (hash << 18) - hash - 1;
-  // zero. For example, the above result for 5 is 43d00600.
+  hash = hash ^ (hash >> 31);
-  //
+  hash = hash * 21;  // hash = (hash + (hash << 2)) + (hash << 4);
-  // We map that to an entry index by masking off the high bits which means
+  hash = hash ^ (hash >> 11);
-  // most small integers would all end up in entry zero. That's bad. To avoid
+  hash = hash + (hash << 6);
-  // that, push a bunch of the high bits lower down so they affect the lower
+  hash = hash ^ (hash >> 22);
-  // bits too.
+  return (uint32_t)(hash & 0x3fffffff);
-  //
-  // The specific mixing function here was pulled from Java's HashMap
-  // implementation.
-  result ^= (result >> 20) ^ (result >> 12);
-  result ^= (result >> 7) ^ (result >> 4);
-  return result;
 }
 
 // Generates a hash code for [num].
@@ -382,7 +376,7 @@ static inline uint32_t hashNumber(double num)
   // Hash the raw bits of the value.
   DoubleBits bits;
   bits.num = num;
-  return hashBits(bits);
+  return hashBits(bits.bits64);
 }
 
 // Generates a hash code for [object].
@@ -429,9 +423,7 @@ static uint32_t hashValue(Value value)
   if (IS_OBJ(value)) return hashObject(AS_OBJ(value));
 
   // Hash the raw bits of the unboxed value.
-  DoubleBits bits;
+  return hashBits(value);
-  bits.bits64 = value;
-  return hashBits(bits);
 #else
   switch (value.type)
   {

test/benchmark/map_numeric.dart.skip

@@ -6,17 +6,17 @@ main() {
 
   var map = {};
 
-  for (var i = 1; i <= 1000000; i++) {
+  for (var i = 1; i <= 2000000; i++) {
     map[i] = i;
   }
 
   var sum = 0;
-  for (var i = 1; i <= 1000000; i++) {
+  for (var i = 1; i <= 2000000; i++) {
     sum += map[i];
   }
   print(sum);
 
-  for (var i = 1; i <= 1000000; i++) {
+  for (var i = 1; i <= 2000000; i++) {
     map.remove(i);
   }
 

test/benchmark/map_numeric.lua

@@ -2,17 +2,17 @@ local start = os.clock()
 
 local map = {}
 
-for i = 1, 1000000 do
+for i = 1, 2000000 do
   map[i] = i
 end
 
 local sum = 0
-for i = 1, 1000000 do
+for i = 1, 2000000 do
   sum = sum + map[i]
 end
 io.write(string.format("%d\n", sum))
 
-for i = 1, 1000000 do
+for i = 1, 2000000 do
   map[i] = nil
 end
 

test/benchmark/map_numeric.py

@@ -6,15 +6,15 @@ start = time.clock()
 
 map = {}
 
-for i in range(1, 1000001):
+for i in range(1, 2000001):
   map[i] = i
 
 sum = 0
-for i in range(1, 1000001):
+for i in range(1, 2000001):
   sum = sum + map[i]
 print(sum)
 
-for i in range(1, 1000001):
+for i in range(1, 2000001):
   del map[i]
 
 print("elapsed: " + str(time.clock() - start))

test/benchmark/map_numeric.rb

@@ -2,17 +2,17 @@ start = Time.now
 
 map = Hash.new
 
-for i in (1..1000000)
+for i in (1..2000000)
   map[i] = i
 end
 
 sum = 0
-for i in (1..1000000)
+for i in (1..2000000)
   sum = sum + map[i]
 end
 puts sum
 
-for i in (1..1000000)
+for i in (1..2000000)
   map.delete(i)
 end
 

test/benchmark/map_numeric.wren

@@ -2,17 +2,17 @@ var start = System.clock
 
 var map = {}
 
-for (i in 1..1000000) {
+for (i in 1..2000000) {
   map[i] = i
 }
 
 var sum = 0
-for (i in 1..1000000) {
+for (i in 1..2000000) {
   sum = sum + map[i]
 }
 System.print(sum)
 
-for (i in 1..1000000) {
+for (i in 1..2000000) {
   map.remove(i)
 }
 

util/benchmark.py

@@ -86,7 +86,7 @@ BENCHMARK("for", r"""499999500000""")
 BENCHMARK("method_call", r"""true
 false""")
 
-BENCHMARK("map_numeric", r"""500000500000""")
+BENCHMARK("map_numeric", r"""2000001000000""")
 
 BENCHMARK("map_string", r"""12799920000""")